You are here:
Estimation of physicochemical properties for PFAS (Per- and Polyfluoroalkyl Substances) using WebTEST 2.0
Citation:
Martin, T., C. Ramsland, N. Charest, C. Lowe, AND A. Williams. Estimation of physicochemical properties for PFAS (Per- and Polyfluoroalkyl Substances) using WebTEST 2.0. QSAR, Copenhagen, DENMARK, June 05 - 09, 2023. https://doi.org/10.23645/epacomptox.23285189
Impact/Purpose:
N/A
Description:
Per- & Polyfluoroalkyl Substances (PFAS) are a growing environmental concern due to their potential toxic effects and persistence in the environment. The physical and chemical properties that make PFAS persistent and mobile in the environment also make them particularly challenging to analyze and remediate. In this study, WebTEST2.0 was utilized to build models for physicochemical properties for PFAS including water solubility, vapor pressure, and octanol water partition coefficient. WebTEST2.0 is web-based platform for real-time predictions using QSAR (quantitative structure activity relationship) models stored in a PostgreSQL database. WebTEST2.0 can develop QSAR models using methodologies such as random forest, support vector machines, and k nearest neighbors. The underlying datasets were constructed using data from many different publicly available data sources. Global (i.e. using data for all chemical classes) and local (i.e. limited to PFAS) models were developed for each property. Global models were found to predict properties for PFAS better than local models. In addition, models were developed with and without genetic algorithm (GA) feature selection. Models with GA feature selection had similar performance to models without it. The views expressed here are those of the authors and do not necessarily represent the views or the policies of the U.S. Environmental Protection Agency.
URLs/Downloads:
DOI: Estimation of physicochemical properties for PFAS (Per- and Polyfluoroalkyl Substances) using WebTEST 2.0
PRESENTATION.PDF (PDF, NA pp, 2218.267 KB, about PDF)